The AI workspace built around on-device LLMs — not one that treats them as an afterthought. Private. Fast. Entirely yours.
macOS · Apple Silicon / Windows · x64 / 100% open on GitHub
Type a task, pick a model, point it at a folder. Zeraix handles the rest — entirely on your machine.
// no api key. no sign-up. the model is already on your machine
Most AI clients are built for cloud APIs and bolt local models on later. Zeraix inverts that — every layer is vertically optimized for on-device inference.
Hop between local and cloud models mid-conversation — context and memory persist across every switch.
File editing, terminal access and web browsing — a full agent, no tokens metered.
Local conversations never touch a server. No telemetry on your prompts. Nothing to leak.
Zeraix profiles your hardware and recommends models that actually run well.
Install skills from the marketplace, automate recurring workflows, and let your assistant grow with you.
The interface ships in 8+ languages, out of the box.
Browse, download and manage models without leaving the app.
Zeraix is shaped in the open. Join the conversation, share feedback, and be the first to try new releases.